Theory
The parts of an IoT system
An IoT system is not just sensors; it is a chain of parts that together take a measurement from the physical world and turn it into a useful action. Knowing these parts, and how data flows through them, is essential for understanding any IoT (or AIoT) system.
The four main components are sensors, actuators, gateways, and the cloud. This lesson explains each and how they connect. On the smart farm, they form a loop: a sensor reads the soil, the data travels through a gateway to the cloud AI, and a command comes back to open an irrigation valve, an actuator. Sense, transmit, process, act.
At a glance
| Component | Role |
|---|---|
| Sensors | Measure/collect data from the physical world (input) |
| Actuators | Act on the physical world in response (output) |
| Gateways | Collect data from local devices, do light processing, and forward it to the cloud |
| Cloud | Store, process, and analyse the data at scale (and run AI), sending decisions back |
Theory
Sensors and actuators: input and output
Sensors are the input side: they measure something physical, temperature, soil moisture, motion, light, an image, and turn it into data. They are the IoT system's senses.
Actuators are the output side: they act on the physical world in response to a decision, turning on a motor, opening a valve, switching a light, moving a part. They are how the system affects the world, not just observes it.
So a complete IoT loop can both sense (sensors) and act (actuators): measure a condition, decide, and change something in response. On the farm, a moisture sensor senses dryness, and an irrigation valve (actuator) acts to water the crop.
Theory
Gateways and the cloud
Between the local devices and the internet sits the gateway. It collects data from many nearby devices, often does some light local processing (filtering, aggregating), and forwards the data onward, bridging the local device network to the internet and the cloud, and translating between them.
The cloud is remote servers that store, process, and analyse the data at scale, and this is often where the AI runs. The cloud can then send decisions or commands back down to the devices (to an actuator). So the layers stack up: perception (sensors and actuators) at the bottom, network (gateways and connectivity) in the middle, and application/cloud (processing, AI, storage) at the top. Data flows up to be understood, and commands flow back down to act.
Quiz
In an IoT system, what is the role of an actuator?
- To measure data from the physical world
- To act on the physical world in response to a decision (e.g. opening a valve or turning on a motor)
- To store data in the cloud
- To provide internet connectivity
Show the answer
To act on the physical world in response to a decision (e.g. opening a valve or turning on a motor)
An actuator is the OUTPUT side of an IoT system: it acts on the physical world in response to a decision, for example opening an irrigation valve, turning on a motor, or switching a light. Option A describes a SENSOR, which is the input side (measuring physical quantities). Option C describes the CLOUD (storing and processing data). Option D describes a gateway/connectivity role (bridging devices to the internet). Keep the roles distinct: sensors sense (input), actuators act (output), gateways forward, the cloud processes. The actuator is how the system changes the world, not just observes it.
Think first
Why have a gateway at all, rather than sending sensor data straight to the cloud?
Why route data through a local gateway instead of every sensor connecting directly to the cloud? Then tap.
Show the answer
Because a gateway solves practical problems of CONNECTIVITY, EFFICIENCY, and coordination that would be costly or impossible if every tiny sensor tried to reach the cloud on its own. First, connectivity and power: many sensors are small, cheap, low-power devices that cannot afford a full internet connection (like Wi-Fi or cellular) themselves, doing so would drain their batteries and raise their cost. Instead they use short-range, low-power links (like Bluetooth or Zigbee) to talk to a nearby gateway, which has the capable connection to reach the internet, so the gateway acts as a shared on-ramp for many local devices. Second, efficiency: a gateway can do LIGHT LOCAL PROCESSING, filtering out noise, aggregating many readings, or discarding unimportant data, before forwarding, so instead of thousands of raw messages flooding the cloud (using bandwidth and incurring cost), only useful, condensed data is sent. This reduces network load, cost, and latency. Third, translation and coordination: local devices may speak different protocols, and the gateway bridges and translates between the local device network and the internet, presenting a unified path to the cloud; it can also coordinate devices locally and keep them running even if the internet connection drops briefly. Fourth, it is a natural place for EDGE computing (processing near the source, a later topic), enabling fast local responses without a cloud round trip. So the gateway is not redundant middleman; it is the sensible bridge that lets many constrained devices connect efficiently and cheaply, cutting bandwidth, saving power, translating protocols, and enabling local processing. Direct-to-cloud for every sensor would be expensive, power-hungry, and unmanageable, which is exactly why gateways exist. Gather locally, forward smartly, is far better than every sensor going it alone.
Summary
Key takeaways
- An IoT system is a chain of four main components: sensors, actuators, gateways, and the cloud.
- Sensors are the input side: they measure physical quantities (temperature, moisture, motion, images) as data.
- Actuators are the output side: they act on the physical world in response (open a valve, turn on a motor).
- Gateways collect data from many local devices, do light processing, and forward it to the internet/cloud.
- The cloud stores, processes, and analyses data at scale (often running the AI) and sends decisions back.
- The layers: perception (sensors/actuators), network (gateways/connectivity), application/cloud (processing/AI).
- Memory hook: sensors sense, actuators act, gateways forward, the cloud processes, data up, commands down.